From 2004 to 2014 i just used what experts said is best. However the more time i spend with particular chips in difeent projest tho more common sonic features common to chip become clear.
So in short-
DIR-
CS8412 double powersuply) is good DIR
CS8414(double power suply) is almost as good as CS8412
DIR9001 Something sonicaly not right about this chip. Edgy,shouty mids and "PRO" sound.
WM8804 much better than DIR9001, but lacks musicality of CS8412. I cant Hear diferencies I2s vs wm8804 in blind test-its so good technicaly.
AK4413- tought choice between it and cs8414.
DF-
SM5803- quite nice, punchy and dynamycs( including 4x OS mode)
SM5813 suprisingly quite nice(as technicaly it should not be that way) sonicaly NOT IDENTICAL to SM5803(smoother)
SM5842- too smooth with its generation dac chips (PCM63), also need carefull capacitor selection to sound clear and with sharper projection,otherwise muddy and too smooth, velvety sound.
PMD100- best widely available and widely used DF (with condition of double PS and reclocking and dither mode 4). no coloration, sound is interesting and inviting, with enought drama to keep you near audio equipment for hours.
PMD200 (real ones) best non OEM available chip,period.
DXF6001/DXF7001- one of the best OEM filter- sharp and musical at the same time. Hughely better than PMD100( especialy with single PS andin hardware mode)
DF1704- it have only one bonus- midrange texture is quite palpable. but bright, tent sound technicaly with all regulators and capacitors(best combiantion to amke it sound less dry and less" hi-fi" -old motorola 7805 and chemi-con AWJ capaitors)
for many years i saw upgrade kits from PMD100 to df1704. sure, in short listening comparision df1704 is more detailed, impressive. but once installed a reduced time of listenting music n hours can be expected. in high end system its clear downgrade comparing to pmd100
DXF7001 and pmd200both capable of 192kh) remains best digital filter (Wadia, Theta DSP filters is not included as I had no ability swap them to toher in same conditions)
Dac chips
UA20400A voltage mode units- smooth musical and always washed out.especialy with tube output (most of nits used tubed outputs for this chip)
UA20400A current output mode units - sharper more transparent. already very nice. still its matter of taste but its not the best R2r dac chip IMHO.
PCM56p-k clean, neutral, fast its like AD1862 with less bass.
PCM63p-k silky, increased midrange texture, requires carefull capacitors and DF selection not so sound soft. I consider this dac chip best but this dac chip best avoided in DIY as its very easy to spoil sound(same goes with sm5842)
PCM1702p-k a sonics combo of pcm56 and pcm63( insipte it should sound simialr to pcm1704)
PCM1704p-k midrange texture reduced comapring to pcm1702(and much more than pcm63) but hights is more refined, extended than pcm63 (and also than pcm1702)
AD1864N-K very clear and transparent and modern sounding dac chips. If i would not knew I would say its quite new sound.
AD1865N-k simialr to sound but worse(!) than 1864n-k could be production year issues, not sure.
AD1853- always musshy a bit compressed closed in mids but pseudo musical just like Elna Silmic MK1 capacitors or OPA2132/34.
PCM1794a- reminds PCM1704 in some ways. refined in i hights, also good dynamycs without being shouty-stays calm.
AKM4495- even wore than AD1853. AKM4499 less mooth more neutral
CS4398- very extended airy hights and large soundstage, but very synthetical sounding (even with audiophile type of outputs)
those are finding of same chips in modular contructions( where part can be easily swaped. in case pmd200/dxf7001, and pmd100/dxf6001 no swaping but swithing chips in same pcb A/B)
its subjective opinion, fortunately no measurements were madeto confirm this and no animals or plants were harmed during this comaparision tests. hope it helps to someone choose their own chips.
sorry for errors, typing fast and with small keyboard laptop.
So in short-
DIR-
CS8412 double powersuply) is good DIR
CS8414(double power suply) is almost as good as CS8412
DIR9001 Something sonicaly not right about this chip. Edgy,shouty mids and "PRO" sound.
WM8804 much better than DIR9001, but lacks musicality of CS8412. I cant Hear diferencies I2s vs wm8804 in blind test-its so good technicaly.
AK4413- tought choice between it and cs8414.
DF-
SM5803- quite nice, punchy and dynamycs( including 4x OS mode)
SM5813 suprisingly quite nice(as technicaly it should not be that way) sonicaly NOT IDENTICAL to SM5803(smoother)
SM5842- too smooth with its generation dac chips (PCM63), also need carefull capacitor selection to sound clear and with sharper projection,otherwise muddy and too smooth, velvety sound.
PMD100- best widely available and widely used DF (with condition of double PS and reclocking and dither mode 4). no coloration, sound is interesting and inviting, with enought drama to keep you near audio equipment for hours.
PMD200 (real ones) best non OEM available chip,period.
DXF6001/DXF7001- one of the best OEM filter- sharp and musical at the same time. Hughely better than PMD100( especialy with single PS andin hardware mode)
DF1704- it have only one bonus- midrange texture is quite palpable. but bright, tent sound technicaly with all regulators and capacitors(best combiantion to amke it sound less dry and less" hi-fi" -old motorola 7805 and chemi-con AWJ capaitors)
for many years i saw upgrade kits from PMD100 to df1704. sure, in short listening comparision df1704 is more detailed, impressive. but once installed a reduced time of listenting music n hours can be expected. in high end system its clear downgrade comparing to pmd100
DXF7001 and pmd200both capable of 192kh) remains best digital filter (Wadia, Theta DSP filters is not included as I had no ability swap them to toher in same conditions)
Dac chips
UA20400A voltage mode units- smooth musical and always washed out.especialy with tube output (most of nits used tubed outputs for this chip)
UA20400A current output mode units - sharper more transparent. already very nice. still its matter of taste but its not the best R2r dac chip IMHO.
PCM56p-k clean, neutral, fast its like AD1862 with less bass.
PCM63p-k silky, increased midrange texture, requires carefull capacitors and DF selection not so sound soft. I consider this dac chip best but this dac chip best avoided in DIY as its very easy to spoil sound(same goes with sm5842)
PCM1702p-k a sonics combo of pcm56 and pcm63( insipte it should sound simialr to pcm1704)
PCM1704p-k midrange texture reduced comapring to pcm1702(and much more than pcm63) but hights is more refined, extended than pcm63 (and also than pcm1702)
AD1864N-K very clear and transparent and modern sounding dac chips. If i would not knew I would say its quite new sound.
AD1865N-k simialr to sound but worse(!) than 1864n-k could be production year issues, not sure.
AD1853- always musshy a bit compressed closed in mids but pseudo musical just like Elna Silmic MK1 capacitors or OPA2132/34.
PCM1794a- reminds PCM1704 in some ways. refined in i hights, also good dynamycs without being shouty-stays calm.
AKM4495- even wore than AD1853. AKM4499 less mooth more neutral
CS4398- very extended airy hights and large soundstage, but very synthetical sounding (even with audiophile type of outputs)
those are finding of same chips in modular contructions( where part can be easily swaped. in case pmd200/dxf7001, and pmd100/dxf6001 no swaping but swithing chips in same pcb A/B)
its subjective opinion, fortunately no measurements were madeto confirm this and no animals or plants were harmed during this comaparision tests. hope it helps to someone choose their own chips.
sorry for errors, typing fast and with small keyboard laptop.
Sorry but CS8412 and CS8414 are technically not good DIR chips to today's standards. There was a time that not much else existed besides the truly horrible YM3623B, then they were the best.
The lack in musicality is in fact lack of jitter with for instance WM8804 (in worst case half of the jitter but often better). CS8412 adds something that should not be there. That WM8804 is good you noticed when you did a direct comparison with I2S and heard no difference.
But what about today's DIR/DAC chips?
The lack in musicality is in fact lack of jitter with for instance WM8804 (in worst case half of the jitter but often better). CS8412 adds something that should not be there. That WM8804 is good you noticed when you did a direct comparison with I2S and heard no difference.
But what about today's DIR/DAC chips?
Last edited:
Without taking notice of implementation ((non)oversampling, current-voltage conversion, output stage and more...), these comparisons are pretty meaningless.
Hey guys it's just his subjective opinion. Thanks to share elviukai.
I have no complaints about wm8804(but i like cs8412 more), DIR9001 is much worse than CS8412 or WM8804. According to jitter figures DIR 9001 should be superior to CS8412. but it isnt.
The reason of writing this texts was conflicts between subjective (what I and my friends hears) and objective (measured or published data) a technican inside me definately would like to like more DIR9001 than cS8412, but.. I have large colection of variuos stock and modified DAC( from old wadias,spectral mark levinsons to highly modified Merason DAC mk1) which I tesedt with quit a bunch speakers and amplifiers and do this in 3 diferent listening rooms.So its not just match some old chip with another old chip but more tendency. so i thought i would share my findings, and maybe sthey will be valuable to other members who likes select parts with subjective selection in mind.
as I told- its purely subjective opinion. If you find it wrong- its perfectly fine. I will not defend it or explain in details. Life is to short for that.
The reason of writing this texts was conflicts between subjective (what I and my friends hears) and objective (measured or published data) a technican inside me definately would like to like more DIR9001 than cS8412, but.. I have large colection of variuos stock and modified DAC( from old wadias,spectral mark levinsons to highly modified Merason DAC mk1) which I tesedt with quit a bunch speakers and amplifiers and do this in 3 diferent listening rooms.So its not just match some old chip with another old chip but more tendency. so i thought i would share my findings, and maybe sthey will be valuable to other members who likes select parts with subjective selection in mind.
as I told- its purely subjective opinion. If you find it wrong- its perfectly fine. I will not defend it or explain in details. Life is to short for that.
Yeah nothing wrong with opinion but with CS8412 it is technically and measurable inferior where DIR9001 is on paper superior but like you I never liked it. Tried it in various ways and also had locking issues. CS8416 is also worse. At least the Wolfson chips perform technically like they should.
I have never used CS8412 with same regulators(or even transformers)- one transformer and double regulation ( 12V>5V) regulators for digital side and same to "analog" side. I also reclokCS8412 signals with 74sn74. maybe this matters. my only complain with cs8412 that it not works with 96/192khz.
Thanks for sharing.
Synchronous or asynchronous reclocking? If it's asynchronous, what's the frequency?
Synchronous or asynchronous reclocking? If it's asynchronous, what's the frequency?
I would also like to share NOS expierences. I have tried NOS in two dac (one with ad1864/pmd100) other with (pcm1702/sm5813) on had RCRCRC filtering another Sallen Key . in both cases NOS sound is midrange forward, houty unclear and not refined hights and quite flat in terms of depth. it could be that NOS with upsampling to 384-768khz or very steep analog filter could be better.
I have also tried PC oversampling and upsampling which(acording to advices here) should be superior to stand alone chips solution but its was not the case.
I have also tried PC oversampling and upsampling which(acording to advices here) should be superior to stand alone chips solution but its was not the case.
Me neither. Almost always 2 regulators with the antiparallel diodes. Long ago I thought the diodes were exxagerated, omitted them… and had lockup 😉
Last edited:
I would only review these two DIRs from my point of view.
I used CS8414 for years with double regulators (now with PMD100 and PCM1702 and earlier NOS TDA1541).
After 8 years, I decided to replace it with a DIR9001, again the same power supply (2xADP7118) and with a much better result and finally sound. The only difference is that I inserted a 74HC541 between DIR and PMD. I recently replaced the ADP7118 with discrete shunt regulators for 3V3 that I made with the help of Mr. Walt Jung and I must admit that it changes things a lot in a positive direction.
probably we value difeent things and that is actualy good thing. i remooved all LT1963 and similar LDO from all my projects. LDO only helps when designs is smeared, wooly form itself. LDO makes sound "better" now mainly using JRC 78xx series regulators 7805,7812,7818,and 7824(for negative i use 78xx not 79xx) i now have hundreds lt1963,3015, lm 2990, 2940, lm 2937, micrel LDO, etc. again i bought them about reading how they are superior to old outdated 78xx series.
when i replaced cs8414 with dir 9001 in 2012 in spencer d1v3 diy dac i also expierenced "big improovement" resolution become smuch bettter, midrange faster and sound nos wooly. soon after i realised that i do not listen to this dac. nothing is canged- muic is the same, me is the same,systemis the same but no desire to listen this dac. i put back cs8414 and soudn become more involving. but it was never good dac comapring to Mark Levinonsons for example. resolution was smeared, and even when i made i2s input and put df1704 resolution become "nearly as good" as other dac i have. But sound of df1704 is not the best..
Elviukai,
Many thanks for taking the trouble to share your experiences.
On a number of essential points we seem to concur which makes the full list a valuable reference for future use. Bookmarked.
Many thanks for taking the trouble to share your experiences.
On a number of essential points we seem to concur which makes the full list a valuable reference for future use. Bookmarked.
Both regulators must start simultaneously unless the antiparallel diodes help.
Is it that (subjectively) more distortion > less distortion, more jitter > less jitter, more wow/flutter/rumble > less wow/flutter/rumble, less dynamic range > more dynamic range, less bits > more bits, etc.?
For example. here are some processors commonly used in the music production process to "improve" the sound:
https://sonicscoop.com/best-tape-saturation-plugins-market/
For example. here are some processors commonly used in the music production process to "improve" the sound:
https://sonicscoop.com/best-tape-saturation-plugins-market/
Many just prefer sound with colorization. It is that simple sometimes. That it are often the people that measure with their ears is pure coincidence 🙂
Just the pretty standard antiparallel diodes between both + connections on the chip. Needed as if one supply ramps up slower than the other the chip may lock up.
I do not like discortions. Unfortunatelly life is is much more complicated than this. Once again -its my subjective opinion. If it helps, I measured many things in my work untill got conflicts between achieved technicaly measured performance and subjective opinion. Than i thought i need more equipment. I also have a friend who works in audio field 30years, have ~8 scopes, but.. listen to wires. Its more mental thing than scientifical- we all want to feel safe. We all feel safiest when we think we know and can explain everything. This is main discusion point on many threads here.
ts so hard not to open can of worms even trying prevent this. Happy listening and good measurements to all of us 👍
ts so hard not to open can of worms even trying prevent this. Happy listening and good measurements to all of us 👍
it has been said many times i think , measurements doesn't tell you if your gonna like it or not it tells you if what you like is "kinda crap" or not , so there is no real conflicts between subjective opinion and measurement , only a choice to make and to assume 🙂
.
.